FROM ZFS TO OPEN STORAGE

FROM ZFS TO OPEN STORAGE Danilo Poccia Senior Systems Engineer Sun Microsystems Southern Europe 1

ZFS Design Principles • Pooled storage

– Completely eliminates the antique notion of volumes – Does for storage what VM did for memory

• End-to-end data integrity

– Historically considered “too expensive” – Turns out, no it isn't – And the alternative is unacceptable

• Transactional operation

– Keeps things always consistent on disk – Removes almost all constraints on I/O order – Allows us to get huge performance wins 2

Why Volumes Exist In the beginning, • Customers wanted more space, bandwidth, reliability each filesystem > Hard: redesign filesystems to solve these problems well managed a > Easy: insert a little shim (“volume”) to cobble disks together single disk. • An industry grew up around the FS/volume model > Filesystems, volume managers sold as separate products > Inherent problems in FS/volume interface can't be fixed

FS

FS

FS

FS

Volume

Volume

Volume

(2G concat) 1G Disk

Lower Upper 1G 1G

(2G stripe) Even 1G

Odd 1G

(1G mirror) Left 1G

Right 1G 3

FS/Volume Model vs. ZFS Traditional Volumes • • • • •

Abstraction: virtual disk Partition/volume for each FS Grow/shrink by hand Each FS has limited bandwidth Storage is fragmented, stranded

FS

FS

FS

Volume

Volume

Volume

ZFS Pooled Storage • • • • •

Abstraction: malloc/free No partitions to manage Grow/shrink automatically All bandwidth always available All storage in the pool is shared

ZFS

ZFS

ZFS

Storage Pool

4

FS/Volume Model vs. ZFS FS/Volume I/O Stack Block Device Interface • “Write this block, then that block, ...”

FS

Object-Based Transactions • “Make these 7 changes to these 3 objects”

ZFS

• All-or-nothing

• Loss of power = loss of on-disk consistency • Workaround: journaling, which is slow & complex

ZFS I/O Stack

Transaction Group Commit

DMU

• Again, all-or-nothing

Volume

• Always consistent on disk

Block Device Interface

• No journal – not needed

• Write each block to each disk immediately to keep mirrors in sync

Transaction Group Batch I/O

• Loss of power = resync

• No resync if power lost

• Synchronous and slow

• Runs at platter speed

Storage Pool

• Schedule, aggregate, and issue I/O at will

5

ZFS Data Integrity Model • Everything is copy-on-write

– Never overwrite live data – On-disk state always valid – no “windows of vulnerability” – No need for fsck(1M)

• Everything is transactional

– Related changes succeed or fail as a whole – No need for journaling

• Everything is checksummed

– No silent data corruption – No panics due to silently corrupted metadata 6

Copy-On-Write Transactions 1. Initial block tree

3. COW indirect blocks

2. COW some blocks

4. Rewrite uberblock (atomic)

7

Bonus: Constant-Time Snapshots • At end of TX group, don't free COWed blocks > Actually cheaper to take a snapshot than not!

Snapshot root

Live root

8

End-to-End Data Integrity Disk Block Checksums

ZFS Data Authentication

• Checksum stored with data block

• Checksum stored in parent block pointer

• Any self-consistent block will pass

• Fault isolation between data and checksum

• Can't even detect stray writes • Inherent FS/volume interface limitation Data

Data

Checksum

Checksum

• Entire storage pool is a self-validating Merkle tree Address Address Checksum Checksum

Data

Disk checksum only validates media ✔ Bit rot

✗ ✗ ✗ ✗ ✗

Phantom writes Misdirected reads and writes DMA parity errors Driver bugs Accidental overwrite

Address Address Checksum Checksum

Data

ZFS validates the entire I/O path ✔ Bit rot ✔ Phantom writes ✔ Misdirected reads and writes ✔ DMA parity errors ✔ Driver bugs ✔ Accidental overwrite

9

RAID-Z Protection RAID-5 and More

• ZFS provides better than RAID-5/6 availability • Striping uses dynamic widths – Each logical block is its own stripe

• All writes are full-stripe writes

– Eliminates read-modify-write (So it's fast!)

• Eliminates RAID-5 “write hole” – No need for NVRAM

• Single, Double and Triple Parity RAID – RAID-Z(1), RAID-Z2, RAID-Z3

10

Self-Healing Data 1. Application issues a

read. ZFS mirror tries the first disk. Checksum reveals that the block is corrupt on disk.

2. ZFS tries the second

disk. Checksum indicates that the block is good.

3. ZFS returns good data

to the application and repairs the damaged block.

Application

Application

Application

ZFS mirror

ZFS mirror

ZFS mirror

11

ZFS Administration • Pooled storage – no more volumes!

– All storage is shared – no wasted space, no wasted bandwidth

• Hierarchical filesystems with inherited properties – Filesystems become administrative control points

– Per-dataset policy: snapshots, compression, backups, privileges, etc. – Who's using all the space? du(1) takes forever, but df(1M) is instant!

– Manage logically related filesystems as a group – Control compression, checksums, quotas, reservations, and more – Mount and share filesystems without /etc/vfstab or /etc/dfs/dfstab – Inheritance makes large-scale administration a snap

• Online everything 12

Data Compression • Reduces the amount of disk space used • Reduces the amount of data transferred to disk – increasing data throughput > LZW ZFS > GZIP (1-9)

Data Compression 13

Copy-on-Write Design Multiple Block Sizes Pipelined I/O Dynamic Striping Intelligent Pre-Fetch

Architected for Speed 14

Creating Pools and Filesystems • Create a mirrored pool named “tank” # zpool create tank mirror c0t0d0 c1t0d0

• Create home directory filesystem, mounted at /export/home # zfs create tank/home # zfs set mountpoint=/export/home tank/home

• Create home directories for several users

Note: automatically mounted at /export/home/{ahrens,bonwick,billm} thanks to inheritance

# zfs create tank/home/ahrens # zfs create tank/home/bonwick # zfs create tank/home/billm

• Add more space to the pool # zpool add tank mirror c2t0d0 c3t0d0 15

Setting Properties • Automatically NFS-export all home directories # zfs set sharenfs=rw tank/home

• Turn on compression for everything in the pool # zfs set compression=on tank

• Limit Eric to a quota of 10g # zfs set quota=10g tank/home/eschrock

• Guarantee Tabriz a reservation of 20g # zfs set reservation=20g tank/home/tabriz

16

ZFS Snapshots • Read-only point-in-time copy of a filesystem

– Instantaneous creation, unlimited number – No additional space used – blocks copied only when they change – Accessible through .zfs/snapshot in root of each filesystem – Allows users to recover files without sysadmin intervention

• Take a snapshot of Mark's home directory # zfs snapshot tank/home/marks@tuesday

• Roll back to a previous snapshot # zfs rollback tank/home/perrin@monday

• Take a look at Wednesday's version of foo.c $ cat ~maybee/.zfs/snapshot/wednesday/foo.c 17

ZFS Clones • Writable copy of a snapshot

– Instantaneous creation, unlimited number – Ideal for storing many private copies of mostly-shared data – Software installations – Workspaces – Diskless clients

• Create a clone of your OpenSolaris source code # zfs clone tank/solaris@monday tank/ws/lori/fix

18

ZFS Send/Receive (Backup/Restore) • Powered by snapshots – Full backup: any snapshot – Incremental backup: any snapshot delta

– Very fast – cost proportional to data changed

• So efficient it can drive remote replication • Generate a full backup # zfs send tank/fs@A >/backup/A

• Generate an incremental backup # zfs send -i tank/fs@A tank/fs@B > /bck/B-A

• Remote replication: send incremental once per minute # zfs send -i tank/fs@11:31 tank/fs@11:32 | ssh host zfs receive -d /tank/fs 19

ZFS Data Migration • Host-neutral on-disk format – Change server from x86 to SPARC, it just works – Adaptive endianness: neither platform pays a tax

– Writes always use native endianness, set bit in block pointer – Reads byteswap only if host endianness != block endianness

• ZFS takes care of everything – Forget about device paths, config files, /etc/vfstab, etc. – ZFS will share/unshare, mount/unmount, etc. as necessary

• Export pool from the old server old# zpool export tank

• Physically move disks and import pool to the new server new# zpool import tank 20

Cost Effective Performance SSDs are 120X more cost effective

• Enterprise HDD > 180 Write IOPS > 320 Read IOPS > 300 GB

• $ per IOPS: 2.43

• Enterprise SSD > 7,000 Write IOPS > 50,000 Read IOPS > 32GB

• $ per IOPS: 0.02 21

ZFS Turbo Charges Applications Hybrid Storage Pools

• ZFS automatically: > Determines data access patterns

and stores frequently accessed data in a read cache called L2ARC

> Bundles IO into sequential staged

writes for more efficient use of low cost mechanical disks

> Very fast synchronous writes

occur to a very fast SSD pool (ZIL) accelerating applications, such as databases & NFS 22

ZFS Hybrid Storage Pool Sun X4250 Storage Server Example Configuration B

Configuration A

4 Xeon 7350 Processors • 32GB FB DDR2 ECC DRAM • OpenSolaris with ZFS •

(1) 80G SSD Cache Device

(1) 32G SSD ZIL Device

(7) 146GB 10,000 RPM SAS Drives

(5) 400GB 4200 RPM SATA Drives 23

ZFS Hybrid Pool Example Based on Actual Benchmark Results 3.2x

4%

4.9x

2x 11%

Read IOPs

Write IOPs

Cost

Storage Power (Watts)

Raw Capacity (TB)

Hybrid Storage Pool (DRAM + Read SSD + Write SSD + 5x 4200 RPM SATA) Traditional Storage Pool (DRAM + 7x 10K RPM 2.5”)

24

OpenSolaris Storage Projects

25

Universal Storage • DMU is a general-purpose transactional object store • ZFS dataset = up to 248 objects, each up to 264 bytes • Key features common to all datasets • Snapshots, compression, encryption, end-to-end data integrity • Any flavor you want: file, block, object, network

Local

NFS

CIFS

ZFS POSIX Layer

iSCSI pNFS

Lustre

DB

Raw

Swap

Dump UFS(!)

ZFS Volume Emulator

Data Management Unit (DMU) Storage Pool Allocator (SPA) 26

Sun Storage 7000 Unified Storage Systems Bringing Simplicity to Enterprise Storage

● ●

●

●

Sun Storage 7110

8 GB RAM Up to 14x300GB SAS 10Krpm Drives ●Up to 4 TB ●2RU package ● ●

Sun Storage 7210

32 GB and 64 GB options ● Up to 142 TB ●Hybrid Storage Pool I/O Acceleration ● Write Flash/SSD options for higher performance ●

Sun Storage 7310 Cluster*

16 GB (up to 64 GB) per controller ● Up to 96 TB ●Hybrid Storage Pool I/O Acceleration ●Read/Write Flash/SSD options for higher performance ●

Sun Storage 7410 Cluster*

64 GB, 128 GB, and 256 GB options per controller ●Up to 288 TB ●Includes cluster software ●Hybrid Storage Pool I/O Acceleration ●Read/Write Flash/SSD options for higher performance ●

Provides support for unified file and block data protocols, including a rich set of data services *Cluster Optional 27

Full Complement of Storage Software Included with the System at No Additional Cost Data Protocols •NFS v2, v3 and v4 •CIFS •ISCSI •HTTP •WebDAV •FTP/SFTP/FTPS •NDMP v3/v4 •NFS over RDMA (Infiniband)

Data Services •Flash Hybrid Storage Pool •Triple-Parity RAID •Triple Mirroring •RAID-Z (5) •RAID-Z DP (6) •Mirroring •Striping •Active-active Clustering •Remote Replication •Antivirus via ICAP Protocol •Snapshots •Clones •Compression •Thin Provisioning •End-to-End Data Integrity •Multi-Path I/O •Fault Management •Microsoft VSS/Shared Folders •Online Data Migration •NFS Nested Mountpoints

Management •DTrace Analytics •Dashboard •Role-Based Access Control •NIS LDAP & AD Alerts •Phone Home •SNMP •Scripting •Upgrade •Hardware View •Advanced Networking •Workflow Automation •User-level Quotas •Support for Microsoft Management Console •System Configuration Backup/ Restore •System Configuration Import/ Export 28

Providing Unprecedented Storage Analytics •Automatic real-time visualization of application and storage related workloads •Simple yet sophisticated instrumentation provides real-time comprehensive analysis •Supports multiple simultaneous application and workload analysis in real- time •Analysis can be saved, exported and replayed for further analysis. •Built on DTrace instrumentation >NFSv3, NFSv4, CIFS, iSCSI, NFS over RDMA (Infiniband) >ZFS and the Solaris I/O path >CPU and Memory Utilization >Networking (TCP, UDP, IP) 29

Instant Visibility into Key Questions “What is CPU and Memory Utilization?” “How much storage is being utilized?” “How is disk performing? How many Ops/Sec?” “What Services are active?” “Which applications/users are causing performance issues?” 30

FROM ZFS TO OPEN STORAGE Danilo Poccia [email protected] blogs.sun.com/danilop 31 22